Limited energy capacity, physical distance between two nodes and the stochastic link quality are the major parameters in the selection of routing path in the internet of things network. To alleviate the problem of stochastic link quality as channel gain reinforcement based Q-learning energy balanced routing is presented in this paper. Using above mentioned parameter an optimization problem has been formulated termed as reward or utility of network. Further, formulated optimization problem converted into Markov decision problem (MDP) and their state, value, action and reward function are described. Finally, a QRL algorithm is presented and their time complexity is analyses. To show the effectiveness of proposed QRL algorithm extensive simulation is performed in terms of convergence property, energy consumption, residual energy and reward with respect to state-of-art-algorithms.

Fog computing is an emerging research field for providing cloud computing services to the edges of the network. Fog nodes process data stream and user requests in real-time. In order to optimize resource efficiency and response time, increase speed and performance, tasks must be evenly distributed among the fog nodes. Therefore, in this paper, a new method is proposed to improve the load balancing in the fog computing environment. In the proposed algorithm, when a task is sent to the fog node via mobile devices, the fog node using reinforcement learning decides to process that task itself, or assign it to one of the neighbor fog nodes or cloud for processing. The evaluation shows that the proposed algorithm, with proper distribution of tasks between nodes, has less delay to tasks processing than other compared methods.

Classification is a crucial process in data mining, data science, machine learning, and the applications of natural language processing. Classification methods distinguish the correlation between the data and the output classes. In single-label classification (SLC), each input sample is associated with only one class label. In certain real-world applications, data instances may be assigned to more than one class. The type of classification which is required in such applications is known as multi-label classification (MLC). In MLC, each sample of data is associated with a set of labels. Due to the presence of multiple class labels, the SLC learning process is not applicable to MLC tasks. Many solutions to the multi-label classification problem have been proposed, including BR, FS-DR, and LLSF. But, these methods are not as accurate as they could be. In this paper, a new multi-label classification method is proposed based on graph representation. A feature selection technique and the Q-learning method are employed to increase the accuracy of the proposed algorithm. The proposed multi-label classification algorithm is applied to various standard multi-label datasets. The results are compared with state-of-the-art algorithms based on the well-known performance evaluation metrics. Experimental results demonstrated the effectiveness of the proposed model and its superiority over the other methods.

The current study presents an incentive contract model to allocate income to venture projects. In this respect, Venture Capital (VC), as one of the main sources of , financing innovative projects, faces several challenges such as moral hazards, information asymmetry, and interest con icts, often referred to as three agency problems. In addition to the identi, cation of the factors that may a, ect the income of venture projects and elaboration of cost functions, this study presented an optimal incentive contract model from the perspectives of venture capitalists and entrepreneurs. In this model, a venture capitalist, as an active investor, provides entrepreneurs with managerial and training assistance. The results revealed that the higher the initial ability of the entrepreneur was, the less money the venture capitalist would pay for training. Of note, in case the venture contract was not accepted, the wealth that the contract parties would obtain would become an in uential factor in the contract payment function. This model was studied considering the bounded rationality hypothesis and implemented using the Q-learning algorithm. In addition, the results obtained from the Q-learning approach were found to be reasonably convergent with the Nash equilibrium.

Q-learning is a one of the most popular and frequently used model-free reinforcement learning method. Among the advantages of this method is independent in its prior knowledge and there is a proof for its convergence to the optimal policy. One of the main limitations of this method is its low convergence speed, especially when the dimension is high. Accelerating convergence of this method is a challenge. Q-learning can be accelerated the convergence by the notion of opposite action. Since two Q-values are updated simultaneously at each learning step. In this paper, adaptive policy and the notion of opposite action are used to speed up the learning process by integrated approach. The methods are simulated for the grid world problem. The results demonstrate a great advance in the learning in terms of success rate, the percent of optimal states, the number of steps to goal, and average reward.